00:02 <Jack21> i have seen that while rendering the same coder twice (deleting cache) i sometimes get errors and sometimes not

00:03 <InPhase> As for the Microsoft urls, they are added by the lib3mf library.

00:04 <Jack21> well they are outdated

00:04 <InPhase> Seems so.

00:05 <InPhase> Jack21: https://github.com/3MFConsortium/lib3mf/issues :)

00:06 <InPhase> Seems no one noticed yet.

00:06 <InPhase> Feel free to tell them.

00:07 <Jack21> i see i should get an github account if i don't stop complaining ..

00:08 <Jack21> for the clamping i think it needs to be done after each step which would be a performance issue

00:09 <Jack21> (i fixed your code example btw - but moving the geometry is not an option )

00:11 <Jack21> but as rotate_extrude is sensetive for the y - it would make sense (and fix the range values they are confusing )

00:17 <Jack2127> so if only the input is checked for every x vector if it is close to 0 (close like 1.e-14)

00:18 <Jack2127> on the other hand i can write a function that is doing this

00:36 <InPhase> Jack21: Turns out the github accounts are free. :) All the cool kids are doing it.

00:53 <Jack21> inphase: but it is microsoft .. think this will do https://bpa.st/VJFQ

01:11 <InPhase> Jack21: Seems to serve your purpose.

04:13 <fling> hi

04:13 <fling> What is the best way to split my models in chunks?

04:13 <fling> for placing like nut sockets and everything

04:13 <fling> for future assembly

04:17 <InPhase> modules?

04:18 <fling> what modules

04:18 <InPhase> Are you asking how do you separate code into chunks?

04:18 <InPhase> Or are you asking how do you design something that is physically separate pieces?

04:19 <fling> yeah how to separate the model chunks, not code

04:21 <fling> for example I might want to kindof cut the bottom of the model and print it as it is unlikely to change and then to focus on the top part which I will change from time to time and only reprint this one thing

04:21 <InPhase> There's no one solution. It depends on the design, and how you want to assemble it. Basically what you do is visualize each piece as a separate part with shared numerical values specifying the dimensions, you calculate those shared values at the top of the file, and then you separately model each part that will later go together as a module.

04:21 <fling> oh

04:22 <InPhase> fling: Here's an example I did: https://www.thingiverse.com/thing:1692539/files

04:22 <othx> InPhase linked to "Twist-on Paper Towel Holder with Rod by rcolyer" on thingiverse => 3 IRC mentions

04:23 <InPhase> fling: There are two holders, and a rod that prints in two parts and screws together, then the rod goes through the two holders.

04:24 <fling> For example I have this half baked duct of a very schwifty shape

04:24 <fling> the idea to split it in two chunks somewhere in the middle

04:24 <fling> I will have to add nut sockets there and everything

04:24 <InPhase> Or this one, top and bottom piece, then it screws together: https://www.thingiverse.com/thing:1686846

04:24 <othx> InPhase linked to "Quick Change Filament Filter by rcolyer" on thingiverse => 3 IRC mentions

04:28 <dalias> my https://www.thingiverse.com/thing:4848302 shows some ideas for how to split into parts to reduce what needs to be reprinted for iterative design

04:28 <othx> dalias linked to "NEMA-17 1:4 step-up gearbox for remote direct drive by dalia5" on thingiverse => 1 IRC mentions

04:29 <fling> I want some piece of code to automagically split general models in half and placing m3 holes properly haha

04:30 <dalias> well you can use CSG to cut them, but it won't really auto-place holes

04:31 <dalias> there's *kinda* a way you can auto-place holes via operations on a 2d projection

04:32 <dalias> https://github.com/richfelker/brickify demonstrates a related trick

04:34 <fling> thanks

08:03 <Jack21> fling: have you worked with children()? so you can make a module that cuts your object in half and places m3 holes and screw holes or other interlocking geometry like dovetails

08:16 <Jack21> fling: i made you an example here https://bpa.st/USMA

09:10 <fling> ah forgot about children

09:10 <fling> thanks

16:05 <gbruno> [github] t-paul closed issue #3927 (script translation-update.sh creates "command not found" error on MacOS (and probably Windows)). https://github.com/openscad/openscad/issues/3927

16:05 <gbruno> [github] t-paul pushed 1 modifications (Merge pull request #3929 from rcolyer/translationscript

16:55 <Guest5315> Howdy! I'm using OpenSCAD on Debian, and need to be able to create PNG files that have the same dimensions for feeding into video clips for Kdenlive. I found I needed to have the size of the window the same for each image. Adjusting viewport distance, etc. is easy enough. I'd like to make the window size for any given instance of OpenSCAD

16:55 <Guest5315> consistent with any other instance. I'm setting my Viewport distance at 300.00, but would like to set the numbers in parentheses to (900x600) for all instances. I suspect this is obvious, simply because I can't seem to find it. Any help here?

16:56 <buZz> Guest5315: you can render to png from commandlines too

16:57 <buZz> there's many commandline options anyway

17:06 <Guest5315> Thank you, I'll take a closer look at those.

23:10 <peeps[zen]> ccox: turns out we can get an easy 35% speedup by using a custom malloc implementation. I just tried this one https://github.com/microsoft/mimalloc

23:11 <peepsalot> I tested on Examples/Old/example024.scad (menger sponge) using the dynamic override option on Linux https://github.com/microsoft/mimalloc#dynamic-override

23:11 <peepsalot> ~28s before, 18s after

23:12 <peepsalot> teepee: this seems like a worthwhile thing to bake in to the build if we can

23:16 <peepsalot> we could also try some other malloc implementations and see if any are better for our particular usage, but mimalloc's performance tests show they beat basically all the other common implementations in *their* benchmarks

23:39 <InPhase> peepsalot: So my suspicious flag has been tripped. The default allocators already get a lot of optimization attention. Usually outperforming means for a specific usage pattern or use case. Did you see any mention elsewhere of what the downside of this one is? (I didn't find it in the readme.)

23:41 <peepsalot> i mean... it used 354.1MB vs 350.8MB memory, so maybe that? well worth it imo

23:42 <peepsalot> not sure if any other downsides